Improving the DSI-II Pro Camera

Looking around the web I've seen many people modifying camera's in the Meade DSI series with good results. The best thing you can do for one of these camera's is deep cool the CCD to lower the noise floor...

I decided on a few other critical goals for the camera upgrade:
1. Low cost
2. Metal working limited to equipment on hand (Lathe/drill press)
2. Peltier cooled with temperature regulation.
3. A sealed housing with descient to absorb moisture and stop condensation.
4. Integrated four position motorized filter wheel.
5. Computer controlled with USB interface.

To control the Peltier, cooling fan, and filter wheel I used an Arudino Duemilanove.
In addition to the above micro-controller, circuitry was also needed to control the stepper motor that moves the filter wheel (a dual "H-Bridge" stepper motor driver) and for temperature control (MOSFET drivers for the Peltier and Fan speed) and an IC to measure the temperature of the cold-finger.

The basic design is shown below:

Image of DSI camera housing

Image of DSI camera schematic

Initial design problems:

Electrical noise:

One design decision was to use MOSFET transistors for switching the high current loads of the Peltier cooling element. Quality MOSFETS can have very low on resistance and are well suited to this task. They are, however, used in an rapidly switching all-on/all-off fashion to control the cooling effort and this can lead to electrical noise. This noise only became noticeable in exposures of >1 or 2 minutes while the CCD was deeply cooled and then only with the DSI II PRO. The DSI I PRO was not as sensitive to it. To suppress the noise I had to add a grounded aluminum shield over the DSI camera board.


Since I had detailed plans for all of the parts in my CAD system, I was presented with only a few surprises. First, the filter wheel placement inside would not allow for positioning the filters directly over the CCD sensor. This turned out to be an non-issue as the clear optical path was still sufficient to fully illuminate the CCD. Secondly, the support screws for the arbor, about which the filter wheel rotates, ended up being under the T-ring adapter that gets screwed down to the outside of the project box. So I had to carefully recess the screw heads to allow for this. And finally, wiring. The Audrino Duemilanove is a bit big for this project case and by the time you're done with USB connectors, ribbon cables between the controller, driver circuit board, and the stepper motor, etc. things are really tight. Had I known, I would have used a smaller Arduino (like a Nano) and that would have been better, but it all fits and works well so I'm leaving it alone.

After thoughts:

An electrical connection to monitor the fan speed would have been nice. If the fan stops the peltier cooler will overheat, not a good situation. I'm pretty sure that this could be detected by looking at the temperature of the cold-finger in relation to the cooling effort and I'll have to take a look at a software solution via that method in the future.
Another useful addition to the camera would be some sort of index system for the filter wheel. An hall effect sensor to zero position the filter wheel would be easy to implement and keep me from having to look at the filter position occasionally. I figure that this can be added when the need arises.
Finally, a bit of circuitry to monitor the 12V supply voltage would be nice too. Then I could keep the controller from trying to operate the filter wheel and Peltier when the voltage for those components isn't within the correct range.

Software features:

Camera Control Software v1.0.0:

The computer connects to the DSI I or DSI II series camera in the usual way. Additionally, the USB interface for the Arduino is connected and appears as an virtual serial port on the user's computer. From there, the user can interact with the Camera body through an Serial console (like Putty) using a simple command set. A detailed command description is on the Camera Driver/Camera Firmware page.

For temperature control the software implements an modified PID control algorithm with an hard coded rate control. This limits the rate of cooling or warming to ease thermal stress on the CCD during use. This means the camera won't cool down quite as fast as would otherwise be possible, but from what I've read this is the way to go.

The set-point and filter position is saved between power cycles to make the camera as automatic as possible.

Sky Planetarium:

Imaging with this camera has been integrated into Sky Planetarium. It now supports imaging with the monochrome DSI I and II PRO (and possibly III PRO) camera's (and also any VFW compatible camera, including the SC modified Phillips ones) along with control of ASCOM compliant filter wheels. Control and reporting of the camera's temperature takes place through my custom ASCOM filter-wheel driver with an extended interface.

Sky Planetarium can now, for example, automatically take a sequence of a given number of images (FITS, BMP, GIF, JPG) of a given duration through the filters you specify. Each image is saved with a filename describing the exposure number and filter used. This refers to the development version which is still a work in progress and will be released once a few more features/fixes are implemented and tested (status as of 5-20-10)


Routinely reaches ΔT's of about 30°C from ambient (as measured by the DSI II's on-board sensor, ΔT > 35°C at the cold-finger)
Temperature regulation to +/-0.1°C or better
No dew problems
Filter wheel functions well, requiring just a few seconds to arrive at any position.
Filter wheel maintains it's position well over many cycles dispite the lack of an index system.

Here are some darks to show the effect that cooling has on the dark noise. These were taken with a DSI II PRO, the histrogram stretch was set for a range of 0 to 10000 for both images:

Before (24.5°C, 1 minute exposure):
Dark frame, before cooling

After ( -5.0°C, 1 minute exposure):
Dark frame, after cooling

4-13-12 Update:

This modified camera has taken some pretty good images, but still wasn't performing up to it's potential...

The camera's MOSFET transistors (for the Peliter cooler, Fan) were poor choices. When the camera is deeply cooling the CCD I can feel the case warming up on the side where they are located. Extra heat in the case is generally a bad thing. I bought myself a little more room inside the case with smaller (cooler running) MOSFETs.

When I originally built the camera the filter wheel wouldn't clear the standoffs (for the DSI Camera board) where it should have been positioned. So, I made a compromise and offset the wheel to allow for this. That was a poor decision and I corrected it this time by making standoffs with an offset to get them out of the way. Then I repositioned the filter wheel to where it should have been to put the filters (centered) under the CCD.

Another problem with the filter wheel was with it's rotation. The camera uses a stepper motor driving a belt around the filter wheel to turn it. I was hoping that I could just count steps and still keep the filters positioned under the CCD. That did not work, it drifts over time and the filters don't end up where they should be. It's a major pain guessing which filter is where when trying to image. So, I added four magnets to the wheel for indexing. Then, I added a hall-effect sensor to the microcontroller and mounted it down in the case so that it signals the arrival of each filter in the correct position. One of the four magnets is set in opposite polarity and provides a "0" index.

Finally, I modified the camera firmware to make it move the filters into place using feedback from that hall-effect sensor. Routines were added to the camera firmware to allow it to calibrate the hall-effect sensor. It spins the filter wheel around and records the various readings to determine the nominal (unaffected) reading from which the North and South magnetic (affected) reading can be detected. About half of the firmware of the camera was rewritten and more control commands are now available for the camera. In operation the camera now puts the requested filter right in the exact position every time. Each time filter 1 is requested the camera spins the filter wheel to find the index magnet (the one with opposite polarity). When filters 2-4 are requested, they are positioned by counting the number of marker magnets that pass as the wheel spins to the correct location. The camera now positions the filters with very high accuracy, and has a positive means of determining which filter is under the CCD.

That's why I like this hobby, I get to tinker with stuff and see how cheaply things can be done... This update cost $16: the magnets were $7, the standoffs were $3, and the new MOSFETs were $6.